March 9, 2024

Man vs Machine - Computer Chess

Computers are very clever machines. They follow sets of coded instructions called ‘programs’ to carry out lots of different tasks. They control traffic lights, read the barcodes on your shopping and even send emails around the world. New programs are making them even more intelligent, but can a computer ever be as smart as (or smarter than) a real person?

Good question.

Artificial intelligence is our name for the ability of computers to carry out the learning and problem-solving tasks that are normally done by the human brain.

One way of testing this intelligence is to create a computer that can play a game and win against the very best human player of that game. A perfect test for this is a game of strategy and skill, such as chess.

BORING. Chess is super boring. Everyone knows that.

Don’t be so quick to judge. Chess is great for your brain – playing it helps to develop problem-solving and decision-making functions. In short, it actually makes you cleverer!

Chess is played by two players who take turns moving their pieces around a chequered board until one of them captures their opponent’s ‘king’. You might have played it already. Chess is believed to have originated in India around 1,500 years ago, but it travelled all over the world and evolved into the game that we play now.

Alan Turing

Almost as soon as people were building modern computers, they were trying to teach them how to play chess. In 1950 the mathematician Alan Turing wrote a theoretical computer program for playing chess. Just six years later the very first chess program was created on an IBM computer and, in 1967, a computer competed against humans in chess tournaments.

Deep Blue Chess Computer

In the mid-1990s, a team of computer programmers at IBM set out to build a computer that they hoped could defeat the world chess champion. They named the computer Deep Blue. It was much larger than anything that you might use at home or in the library. It was 6 feet, 5 inches (2 metres) tall, weighed 1.4 tonnes and it used 256 processors! A processor is the chip inside the computer that performs calculations. These processors made Deep Blue the world’s most powerful chess computer searching 200 million chess positions per second. A human brain can only search two or three moves in that time, so the computer had a huge advantage when it came to speed. But was it as good at playing the game?

In order to play chess, Deep Blue was programmed to use a mathematical model of game theory called a game tree. The computer uses this ‘tree’ to calculate a sequence of moves that leads to the best possible path to win the game. It is called a game tree because the model looks like a collection of branches. The number of different ways in which the game can be played makes the tree bigger, so a game tree for chess would be enormous. The computer then uses mathematical instructions called algorithms to reduce the size of the game tree and find the best moves more quickly.

Garry Kasparov

Deep Blue’s human opponent was the chess grandmaster and champion Garry Kasparov. He had played the game since he was a little boy and he had become the youngest ever world champion when he was just 22. He was the perfect opponent for this new computer.

In February 1996, Kasparov and Deep Blue played a six-game chess match. Deep Blue won the first game, but Kasparov went on to win three. The final two were draws. Deep Blue had only won one game, and it was clear that the human brain was able to defeat the computer.

Just over a year later there was a rematch. Again they played six games. Three of the games were draws, but this time Deep Blue beat Kasparov twice. The computer was declared the winner. The machine had beaten the man. Deep Blue was retired, and parts of it are now on display in American museums.

Whatever happened to winner stays on? Ed

How artificially intelligent was Deep Blue?

All of its abilities were programmed into it by the team at IBM with the help of a chess grandmaster named Joel Benjamin. Deep Blue could only play what it had been taught. It was capable of searching the game tree very quickly but the computer was unable to learn from the way in which Kasparov played.

Machines are now developing better artificial intelligence. The newest computers are tested by ‘Go’, a game which is even older and even more complicated than chess. A Google computer named AlphaGo repeatedly beat the best Go players in the world between 2015–2017. Unlike Deep Blue, AlphaGo learnt as it played, and improved without the help of programmers.

It is hoped that artificial intelligence will soon be able to drive our cars and improve our medical care. By competing in games like chess and Go, computers are becoming able to ‘learn’ and ‘think’. They will soon work alongside humans in a very exciting future.

Extra facts:

IBM stands for International Business Machines Corporation and the company was first founded in 1911.

As long ago as 1846, a mathematician named Charles Babbage came up with the idea for a computer which could play noughts and crosses.

IBM built computers that NASA used to work out the calculations for sending rockets into space in the 1960s.

The ‘Turing Test’ is a test designed by the mathematician Alan Turing to check whether a computer can think in the same way as a human.

An IBM processor is currently working on Mars as a part of the Curiosity Rover.

In 2011, a computer named Watson was victorious against humans on a television game show called ‘Jeopardy!’.

Deep Blue’s two towers are now on display in the Computer History Museum in California and the Smithsonian Museum in Washington D.C.

If you enjoyed this interesting blog, then you will love the articles in our AI issue. This article appeared in the Game Theory issue and which you can access via our digital subscription!

Click here to join AQUILA today for more!

Words: Frances Durkin. Illustration: Yann Bastard